情報処理学会 第80回全国大会講演要旨

This study aims to investigate the underlying neural mechanism of phantom noise in autism spectrum disorder (ASD) from a constructive approach based on a computational model. Previous experiments on the people with ASD showed that the phantom noise may occur in an unexpected environment. We hypothesize that both top-down prediction and bottom-up influence from spontaneous neural activities are involved in perceiving sensory stimuli, and that in case of individuals with ASD, the deficit of prediction ability might cause phantom noise. We model a source of spontaneous neural activities as stochastic resonance. The experiment manipulated the time window of referred preceding signal to examine how it affects hyper neural activities. The computer simulation results support our hypothesis that impaired prediction from the preceding signal could produce phantom noise in case of auditory perception.